RASA Training Model

Rasa Open Source
1

Hi ,

I am trying to do below code but facing an error while execution.

Not able to fix it. Please help

Code -

import asyncio

fallback = FallbackPolicy(fallback_action_name=“utter_unclear”, core_threshold=0.2, nlu_threshold=0.5)

agent = Agent(‘domain_ad.yml’, policies=[MemoizationPolicy(max_history=2), KerasPolicy(epochs=500, batch_size=10)])

loading our neatly defined training dialogues

training_data = agent.load_data(‘stories_ad.md’)

loop = asyncio.get_event_loop()

loop.stop()

loop.close()

training_data = loop.run_until_complete(agent.load_data(‘stories_ad.md’)) agent.train(training_data)

loop.close()

agent.train( training_data, validation_split=0.0 )

agent.persist(‘models/nlu/helpdesk’)

Issue-

RuntimeError: This event loop is already running

Processed Story Blocks: 100%|█████████████████████████████████████████████| 8/8 [00:00<00:00, 640.95it/s, # trackers=1] Processed Story Blocks: 100%|█████████████████████████████████████████████| 8/8 [00:00<00:00, 617.07it/s, # trackers=8] Processed Story Blocks: 100%|████████████████████████████████████████████| 8/8 [00:00<00:00, 135.93it/s, # trackers=50] Processed Story Blocks: 100%|████████████████████████████████████████████| 8/8 [00:00<00:00, 138.29it/s, # trackers=50]

Can Anyone tell me why this is coming and what i have to do stop it ?

Because of this issue my model is not getting saved in the model directory.

RASA version -1.7.0

2

Hi @Aditya07! Can you please share your config and stories? Out of curiosity, why did you write this code instead of just using the Rasa CLI?

3

Sure I will share.

I am using Jupyter NB to make my chatbot. Would you share how to use CLI and also how i can use RASA with Jupyter NB.

I am new to RASA so not know lot of things

stories_ad.md (508 Bytes) config_adi.yml (446 Bytes)

4

One more issue now - After completing 500 epochs , Python throws the below error. I have no clue how to solve this.

@tyd Please help

from rasa.core.policies.fallback import FallbackPolicy

from rasa.core.policies.keras_policy import KerasPolicy

from rasa.core.policies.memoization import MemoizationPolicy

from rasa.core.agent import Agent

​

import asyncio

​

​

fallback = FallbackPolicy(fallback_action_name = “utter_unclear”,

core_threshold = 0.2,

nlu_threshold = 0.5)

​

agent = Agent(‘domain_ad.yml’, policies = [MemoizationPolicy(max_history = 2),

KerasPolicy(epochs = 500, batch_size = 10)])

​

# loading our neatly defined training dialogues

# training_data = agent.load_data(‘stories_ad.md’)

​

​

loop = asyncio.get_event_loop()

# loop.stop()

# loop.close()

training_data = loop.run_until_complete(agent.load_data(‘stories_ad.md’))

agent.train(training_data)

# loop.close()

agent.train(

training_data,

validation_split = 0.0

)

​

agent.persist(‘models/nlu/helpdesk’)

Issue –

---------------------------------------------------------------------------

ValueError Traceback (most recent call last)

~.conda\envs\python-cvcourse\lib\site-packages\tensorflow_core\python\client\session.py in _run (self, handle, fetches, feed_dict, options, run_metadata)

1119 subfeed_t = self.graph.as_graph_element(

-> 1120 subfeed, allow_tensor=True, allow_operation=False)

1121 except Exception as e :

~.conda\envs\python-cvcourse\lib\site-packages\tensorflow_core\python\framework\ops.py in as_graph_element (self, obj, allow_tensor, allow_operation)

3606 with self . _lock :

-> 3607 return self . _as_graph_element_locked ( obj , allow_tensor , allow_operation )

3608

~.conda\envs\python-cvcourse\lib\site-packages\tensorflow_core\python\framework\ops.py in _as_graph_element_locked (self, obj, allow_tensor, allow_operation)

3685 if obj . graph is not self :

-> 3686 raise ValueError ( “Tensor %s is not an element of this graph.” % obj )

3687 return obj

ValueError : Tensor Tensor(“Placeholder:0”, shape=(), dtype=float32) is not an element of this graph.

During handling of the above exception, another exception occurred:

Please consider the environment before printing.

5

@Aditya07 The best way to train is to use the CLI (e.g. rasa train). Where did you find this code?

6

Sure Thanks will try CLI

Code was found from one of github repo